The Relationship between Relevance and Reliability in Internet-based Information and Retrieval Systems

Daniel E. O 'Leary

In this paper, interact information retrieval systems are modeled as an intermediate step between users of the system and the original or expected information. For example, use of a search engine will generate a "report" providing a list of URLs and a corresponding brief description that may or may correctly describe the label being searched. In particular, there is a probability that the report of the content (x#) and the actual content (x) are the same, where Pr(x#1x) is used to define the reliability of the system. If reliability is not considered in interact systems, non optimal behavior may be promulgated in both the design and use of those systems. This suggests that system reliability is an important issue to be studied. The basic model of information relevance in the information retrieval process is reviewed, where the precision is a function, in part, of the recall and fallout rate. Then that relevance model is extended to include reliability due to the intermediate nature of the system and the implications of that reliability are studied. Reliability is found to have a major impact on precision and fallout rates. Thus, decisions on the design of intemet information and retrieval systems also are impacted.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.